Incoming Mail Attachments

[Joshua Smith]

The attachments property on incoming mail is a list of pairs: name,
contents

(note that the docs say "attachments is a list of element pairs
containing file types and contents." which is not correct)

The contents are still encoded. For example:

From nobody Mon Oct 19 15:47:38 2009
content-transfer-encoding: base64

iVBORw0KGgoAAAANSUhEUgAAAaUAAAEUCAIAAAAEJr4pAAAXrGlDQ1BJQ0MgUHJvZmlsZQAAeAHt
...

Do I need to write something to parse this and deal with the different
possible values of content-transfer-encoding? Or is there a built-in
or an appengine API that will do that for me?

Also, when there are references to these attachments within the HTML,
such as
<img src="cid:b2ea907c-e731-43ff-a932-ba307782ba34" >
it appears that there is no way to know what the mapping is between
the cid's and the attachments.

This makes it impossible for me to display the email in a browser with
the images in-line.

-Joshua

[Joshua Smith]

OK, I've made a little progress on this. It turns out that the
attachments member is using a EncodedPayload object to hold the
contents.

This has the delightful little function decode() which *almost* give
you the raw data. However, there is a line in mail.py that goes:

payload = payload.decode(self.encoding).lower()

why does it say "lower()" at the end? No idea, but it completely
trashes the data.

-Joshua

[Rafe]

The content is in fact an instance of EncodedPayload. You can
decode the content by calling 'decode' on the EncodedPayload object.

Hope that this is what you were looking for.

- Rafe Kaplan

[Joshua Smith]

Yes, it was, although there is apparently a bug in decode() in that it
lower-cases everything!

This bug affects not only attachments, but also HTML bodies!

Here's the work-around:

for a in message.attachments:
name = a[0]
payload = a[1].payload.decode(a[1].encoding)
... do something with name and payload ...

[Joshua Smith]

I've reported the bug:

Issue 2289

Here is a more complete work-around:

def goodDecode(encodedPayload):

  encoding = encodedPayload.encoding

  payload = encodedPayload.payload

  if encoding and encoding.lower() != '7bit':

    payload = payload.decode(encoding)

  return payload

usage:

    bodies = message.bodies(content_type='text/html')

    allBodies = "";

    for body in bodies:

      allBodies = allBodies + "\n" + goodDecode(body[1])

and:

      if hasattr(message, 'attachments'):

        for a in message.attachments:

          msg.attachmentNames.append(a[0])

          msg.attachmentContents.append(db.Blob(goodDecode(a[1])))

        msg.put()

On Oct 19, 2009, at 3:34 PM, Rafe wrote:

[Joshua Smith]

I've created issue 2291 for the cid: problem.

On Oct 19, 2009, at 1:07 PM, Joshua Smith wrote:

[Joshua Smith]

Got a notice that this was fixed.  I presume that means it will be fixed in the next SDK update.